Issue description
In the attached screen there is an extract from scoring code generated by sasctl for Python models when imported into SAS Model Manager.
The key part is an option: dtype=object which is probably fine for sci-kit learn based models but it causes the errors for LGBoost and XGBoost models.
When any of these package is used to build and import the model into SAS Model Manager the scoring test will fail due to PyMas execute function error: DS2 "pymas" package encountered a failure in the 'execute' method
ScoringTest.log
In CAS log You can find the root cause for that error:
Extract sas-cas-server.log
The object data type is not supported by such popular Python models. I had to manually remove dtype=object , which means the data type will be inferred. And then the Test Score works fine.
I hope the intention of sasctl package is support Python developed models, not only sci-kit learn based ones?
Issue description In the attached screen there is an extract from scoring code generated by sasctl for Python models when imported into SAS Model Manager. The key part is an option: dtype=object which is probably fine for sci-kit learn based models but it causes the errors for LGBoost and XGBoost models. When any of these package is used to build and import the model into SAS Model Manager the scoring test will fail due to PyMas execute function error: DS2 "pymas" package encountered a failure in the 'execute' method ScoringTest.log In CAS log You can find the root cause for that error: Extract sas-cas-server.log The object data type is not supported by such popular Python models. I had to manually remove dtype=object , which means the data type will be inferred. And then the Test Score works fine. I hope the intention of sasctl package is support Python developed models, not only sci-kit learn based ones?
Version 1.10.1